MathTutorResearch

Research on Large Language Model (LLM) capabilities in mathematics tutoring and step-by-step reasoning.

🔬 Research Focus

This project investigates how well current LLMs can:

1. Solve complex mathematical problems accurately.
2. Explain their reasoning in a pedagogical, step-by-step manner suitable for tutoring.
3. Correct student errors by identifying specific misconceptions.

📂 Content

The repository contains Jupyter Notebooks documenting experiments with various prompts, model configurations, and datasets.

• Prompt Engineering: Techniques to encourage 'Chain of Thought' reasoning.
• Error Analysis: Systematically categorizing where models fail (arithmetic vs. logic).
• Fine-Tuning: Exploration of fine-tuning strategies to improve mathematical reasoning.

🛠️ Tools Used

• Python: For scripting and data analysis.
• Jupyter Notebooks: For documenting experiments and results.
• OpenAI API / Hugging Face: Accessing state-of-the-art LLMs.
• LangChain: Orchestrating complex model interactions.

🚀 Usage

Clone the repo and run the notebooks to see the analysis and results:

git clone https://github.com/yashmahe2020/math-tutor-research.git

---

A deep dive into the intersection of AI and Education.